COS 136-4
Empirical succession mapping and data assimilation to constrain demographic processes in an ecosystem mode

Friday, August 14, 2015: 9:00 AM
318, Baltimore Convention Center
Ryan Kelly, Department of Earth and Environment, Boston University, Boston, MA
Travis Andrews, Earth & Environmental Sciences, Lehigh University, Bethlehem, PA
Michael Dietze, Earth and Environment, Boston University, Boston, MA
Background/Question/Methods

Shifts in ecological communities in response to environmental change have implications for biodiversity, ecosystem function, and feedbacks to global climate change. Community composition is fundamentally the product of demography, but demographic processes are highly simplified or missing altogether in many ecosystem, Earth system, and species distribution models, which together account for most efforts to forecast future ecological change. This limitation arises in part because demographic data are noisy and difficult to synthesize. As a consequence, demographic processes are challenging to formulate in models in the first place, and to verify and constrain with data thereafter. Here, we used a novel analysis of the USFS Forest Inventory Analysis to improve the representation of demography in an ecosystem model. First, we created an Empirical Succession Mapping (ESM) based on ~1 million individual tree observations from the eastern U.S. to identify broad demographic patterns related to forest succession and disturbance. We used results from this analysis to guide reformulation of the Ecosystem Demography model (ED), an existing forest simulator with explicit representation of demography.

Results/Conclusions

Results from the ESM reveal a coherent, cyclic pattern of change in temperate forest tree size and density over the eastern U.S. The ESM captures key ecological processes including succession, self-thinning, and gap-filling, and quantifies the typical trajectory of these processes as a function of tree size and stand density. Recruitment is most rapid in early-successional stands with low density and mean diameter, but slows as stand density increases, and mean diameter increases until thinning promotes recruitment of small-diameter trees. Strikingly, the upper bound of size-density space that emerges in the ESM conforms closely to the self-thinning power law often observed in ecology. By comparison, the ED model obeys this same overall size-density upper bound, but simultaneously overestimates plot-level growth, mortality, and fecundity rates, leading to unrealistic emergent demographic patterns. In particular, the current ED formulation cannot capture steady state dynamics evident in the ESM. Ongoing efforts are aimed at reformulating ED to more closely approach overall forest dynamics evident in the ESM, and then assimilating inventory data to constrain model parameters and initial conditions.